Bearings in which a plurality of balls are disposed at circumferentially spaced locations between the raceways of respective inner and outer rings are well-known in the art. Generally, such bearings employ retainers, or cages, which are disposed between the inner and outer rings at a spacing therefrom for the purpose of maintaining the desired circumferential spacing between the bearing balls. To facilitate assembly of the ball bearing with the balls between the inner and outer rings, these cages are generally formed in two pieces which mate along an interface normal to the axis of the bearing. In the past, side plates, rivets and the like have been used to secure the cage halves together after the bearing balls have been placed in the ball pockets formed in the cage. Not only do such elements entail additional complexity and expense, but such rivets are susceptible to failure, reducing the useful life of the bearing.
Farley et al U.S. Pat. No. 4,451,098 discloses a ball cage formed as two portions, one of which is formed with circumferentially spaced fingers between the ball pockets which engage portions of the mating cage portion to hold the cage portions together. While such an assembly does not require side plates or rivets to hold the cage portions together, it nevertheless has several disadvantages. First, the shape of the ball pockets and retaining fingers is such as to result in a relatively complicated and hence expensive process of fabrication. Second, the disclosed assembly apparently does not permit relative rotational movement between the cage portions. Such movement has been found desirable in preventing the balls from locking in their pockets. Such locking increases bearing torque and decreases the stability of the cage.